Download paper

Comitative and Plural Pronoun Constructions
Anna Feldman
Department of Linguistics, The Ohio State University
[email protected]
1
Introduction
Russian has two different means of combining two NPs to form a plural NP.
(1)
[i-conjunction]
i
Daša
xodjat v školu.
(a) Maša
Masha.Nom and Dasha.Nom go.PL to school
Masha and Dasha go to school.
(b) [s-conjunction]
Maša
s
Dašei
xodjat v školu.
Masha.Nom with Dasha.Instr go.PL to school
Masha and Dasha go to school.
The construction in (1a) and in (1b) mean exactly the same and trigger exactly the same
type of agreement with the verb. The question that I want to solve in this paper is whether
(1a) and (1b) have the same syntactic structure.
However, s- adjuncts do not trigger the plural agreement with the verb:
(2)
[s-adjunction]
Anja s
Vanjei pošla
v biblioteku
Anya with Vanya went.SG.Fem to library
Anya went to the library with Vanya.
In addition, there is another interesting construction in Russian, illustrated in (3) by the
first reading. (The second reading is the instance of s-adjunction, as will be shown below). Some analyses assume that this construction (on the first reading) is also an instance of s-coordination (e.g. McNally (1993)), others ascribe the adjunction structure to it
(e.g.Ladusaw (1989)).
(3)
My
s
Dašei
xodim v školu.
We.Nom with Dasha.Instr go.PL to school
Dasha and I go to school./Dasha and we go to school.
I will begin the discussion with the syntactic and semantic differences between with-conjuncts
and with-adjuncts. In order to make my exposition simpler, I will examine data that involve
the comitative construction with singular noun phrases; however, the same arguments are
applicable to the comitative construction with plural noun phrases. Having established that
there are two types of with-phrases, I will compare s(with)-coordination with i(ordinary)coordination. I will show that the two types of coordination realize different syntactic
structures. Then, I will argue that examples like the first reading of (3) are instances of the
Inclusive Plural Pronoun Construction (IPPC). A formal account of the three comitative
types within the framework of HPSG will be proposed.
2
Comitative Adjuncts vs. Comitative Conjuncts
Russian has comitative prepositional phrases which are VP-adjuncts consisting of the preposition s (with) and an instrumental case-marked NP. I will summarize the main arguments in favor of distinguishing adjunction from s-coordination. Some of these arguments
have been mentioned in previous literature with regard to Polish, Latvian, Hungarian and
other languages (see Schwartz (1988),Schwartz (1988),Ladusaw (1989),Dyła (1988),Aissen (1988),Aissen (1989),Urtz (1994),Vassilieva (2000) among others).
2.1 Number Agreement
In Russian, only nominative-marked NPs agree with the verb. If a sentence lacks a nominativemarked NP, then the verb is in its third person singular neuter form, as can be seen from
the following examples:
(4)
(a) Oni
xotjat
pit’.
They.Nom want.PL drink.Inf
They want to drink.
(b) Im
xočetsja
pit’.
They.Dat want.neuter.SG.refl drink.Inf
They want to drink.
(5)
(a) [adjunction]
Maša
s
Dašei
xodit
v školu.
Masha.Nom with Dasha.Instr go.3SG to school
Masha goes to school with Dasha.
(b) [coordination]
Maša
s
Dašei
xodjat v školu.
Masha.Nom with Dasha.Instr go.PL to school
Masha and Dasha go to school.
In (5a), Maša is nominative and hence require the singular agreement with the verb; however, the verb in (5b) exhibits plural agreement, which seems to indicate that the verb has a
plural subject in the nominative case. This fact indicates that the two sentences have different syntactic structures: (5a) is adjunction, whereas in (5b), the NP-s-NP cluster is in fact
acoordinate phrase. This is not the only argument though.
2.2 Reflexive Pronouns
Another set of facts that seem to indicate that comitatives can be divided into two types—
coordinate and adjunct comitatives—involves reflexive pronouns. Consider the following
set of sentences:
(6)
[adjunction]
prodat’ svoi /*ejo komputer.
Dašei rešila
(a) Maša s
Masha with Dasha decided.SG sell.Inf self-/her
computer
Masha and Dasha decided to sell Masha’s computer.
(b) Maša s
prodat’ *svoi /*ejo
Dašei rešila
Masha with Dasha decided.SG sell.Inf self-/her
Masha and Dasha decided to sell Dasha’s computer.
Dašei rešila
prodat’ svoi* + /*ix +
(c) Maša s
Masha with Dasha decided.SG sell.Inf self-/their
Masha and Dasha decided to sell their computer.
komputer.
computer
komputer.
computer
In (6), the only antecedent for svoi is Maša. The reflexive cannot be bound jointly by Maša
and Daša. In (7), on the other hand, the situation is reversed: the possesive reflexive cannot
be anteceded by Maša alone since it must have plural agreement. Thus, it must be bound
jointly by Maša and Daša.
(7)
[coordination]
Dašei rešili
prodat’ *svoi /*ejo komputer.
(a) Maša s
Masha with Dasha decided.PL sell.Inf self-/her
computer
Masha and Dasha decided to sell her computer.
(b) Maša s
Dašei rešili
prodat’ *svoi /*ejo komputer.
Masha with Dasha decided.PL sell.Inf selfcomputer
Masha and Dasha decided to sell her computer.
(c) Maša s
Dašei rešili
prodat’ svoi + /*ix +
Masha with Dasha decided.PL sell.Inf selfMasha and Dasha decided to sell their computer.
komputer.
computer
Note also that when a prepositional phrase occurs postverbally, the agreement with the verb
is always singular, whereas a prepositional phrase in preverbal position seems to comprise
a constituent together with the NP that triggers plural agreement:
(8)
(a) [adjunction]
Maša rešila
s
Dašei prodat’ svoi komputer.
Masha decided.SG with Dasha sell.Inf self- computer
Masha and Dasha decided to sell Masha’s computer.
(b) [coordination]
*Maša rešili
s
Dašei prodat’ svoi +
Masha decided.PL with Dasha sell.Inf selfMasha and Dasha decided to sell their computer.
komputer.
computer
Thus, the agreement facts suggest that there are two comitative types: adjunction and coordination.
2.3 Discontinuity
If we adopt the hypothesis that comitative constructions are homogeneous and the NP-s-PP
string is one constituent, then the fact that comitative constructions behave differently with
respect to discontinuity becomes puzzling. It seems that ’extraction’ out of the NP-s-NP
string is allowed in the comitative constructions with singular agreement and forbidden in
the comitative coordinate constructions with plural agreement. This fact suggests that the
former type of construction is adjunction and the latter is coordination.
(9)
(a) [adjunction]
S
kem
Maša
pošla
v kino?
With whom-Instr. Masha.Nom went.SG to movie
With whom did Masha go to the movies?
(b) [coordination]
*S
kem
Maša
pošli
v kino?
With whom-Instr. Masha.Nom went.PL to movie
With whom did Masha go to the movies?
(c) [coordination]
*Kto
s
Mašei
pošli
v kino?
Who.Nom with Masha.Instr went.Pl to movie
Who went to the movies with Masha?
(d) [adjunction]
Kto
s
Mašei
pošol
v kino?
Who.Nom with Masha.Instr went.Sg to movie
Who went to the movies with Masha?
If we adopt another hypothesis, namely, that the NP and the s-PP are two separate constituents, then in a language like Russian, which has relatively free word order, there is no
reason to expect (11) to be ungrammatical. However, the ungrammaticality of (11) suggests
that the NP-s-NP string is in fact a single constituent.
(10) Tanja s
Vanei zavtra/nakonets-to
poidut
v restoran.
Tanya with Vanya tomorrow/eventually go.FUT.PL to restaurant
Tomorrow (Eventually),Tanya and Vanya will go to the restaurant.
(11) *Tanja zavtra/nakonets-to
s
Vanei poidut
v restoran.
Tanya tomorrow/eventually with Vanya go.FUT.PL to restaurant
Tomorrow (Eventually),Tanya and Vanya will go to the restaurant.
(12) Tanja zavtra/nakonets-to
s
Vanei poidyot
v restoran.
Tanya tomorrow/eventually with Vanya go.FUT.SG to restaurant
Tomorrow (Eventually),Tanya will go to the restaurant with Vanya.
2.4 Semantic Difference
The meaning of s ’with’ sometimes makes it difficult to distinguish between the adjunct
reading (13a) and the conjunct reading as in (13b):
(13)
(a) [adjunction]
Maša xodit
s
Dašei v školu.
Masha goes.SG with Dasha to school
Masha and Dasha go to school.
(b) [coordination]
Maša s
Dašei xodjat v školu.
Masha with Dasha go.PL to school
Masha and Dasha go to school.
It is hard to put one’s finger on the semantic difference between the comitative adjunction
and the comitative coordination. However, there are ways to disambiguate a construction,
forcing a coordination-only meaning1 . Coordinate constructions can have either a collective or a distributive reading, whereas s-adjuncts can have only a collective reading, as in
the English example: Mary read the book with John. An s-adjunct cannot modify a verb
that demands a distributive reading. Thus, for example, verbs such as believe or know will
not have with-modifiers. However, s-phrases with coordinate meanings are acceptable with
these verbs in (14b) just as are regular coordinated subjects (14c):
(14)
(a) [adjunction]
*Maša verit
s
Dašei
v boga.
Masha believe.3SG.Pres with Dasha.Instr in God
*Masha believes in God with Dasha.
(b) [s-coordination]
Maša s
Dašei
verjat
v boga.
Masha with Dasha.Instr believe.3PL.Pres in God
Masha and Dasha believe in God.
(c) [ordinary coordination]
Maša i
Daša
verjat
v boga.
Masha and Dasha.Instr believe.3PL.Pres in God
Masha and Dasha believe in God.
1
The cases discussed in this section serve as counterexamples to the claim made by McNally (1993) that
comitatives always have a collective reading.
The ability of a s-phrase to occur with verbs like verit’ ’believe’ that only trigger distributive
readings can be considered to be a diagnostic for s-coordination.
In addition, comitative adjuncts cannot occur with reciprocals:
(15)
(a) [coordination]
Maša s
Dašei ljubyat drug druga.
Masha with Dasha like
each other
Masha and Dasha like each other.
(b) [adjunction]
*Maša s
Dašei ljubit drug druga.
Masha with Dasha like each other
*Masha likes each other with Dasha.
Additional evidence in favor of distinguishing the two syntactic structures comes from the
following examples:
(16)
(a) [adjunction]
Maša s
Dašei zarabotala tri
rublja.
Masha with Dasha earned.SG three rubles
Masha earned three rubles with Dasha.
(b) [coordination]
Maša s
Dašei zarabotali tri
rublja.
Masha with Dasha earned.PL three rubles
Masha and Dasha earned three rubles/Masha and Dasha each earned three
rubles (6 rubles total)
As becomes evident from the translations of the examples provided in this section, the
comitative adjuncts do not allow a distributive reading, whereas the comitatives conjuncts
allow the object to either distribute over the subject or not.
Thus, the agreement, binding, extraction and semantic facts clearly show that there are two
types of comitative constructions: adjunction and coordination.
3
NP-NP and NP-PP Coordination Hypotheses
The next question is whether comitative and ordinary coordination have the same syntactic structure. From the data discussed so far, it is natural to assume that ordinary and
s-coordination are the same, since their behavior seems to be identical. However, the existence of a special form of conjoining two NP’s might hinge on the existence of a different
syntactic structure. In this section, I will explore the two possible hypotheses.
3.1 Dyła 1988
One of the most natural assumptions to make with respect to s-coordination is to assume
that it is exactly the same as ordinary coordination, or at least very similar.
Dyła (1988), who considers an analogous phenomenon in Polish, treats comitative constructions that trigger plural verb agreement as ‘quasi-comitative’; i.e. they look like comitative constructions but exhibit the syntactic properties of coordinate constructions. Dyła’s
main argument is the fact that both ordinary and comitative coordination 2 triggers the plural
agreement with the verb. Dyła (1988) argues that comitative coordinate phrases (‘quasicomitative constructions’) are indeed coordinate constructions. Consider the following Polish example Dyła (1988):
(17) Ewa
z
Jankiem poszli
na spacer.
nom.fem with instr.masc went-3pl.masc for walk
Eve and John went for a walk.
This comitative construction is analogous to the Russian examples discussed in the previous
sections. The difference is that in Polish the picture becomes more complicated due to
gender agreement. Only nominative-marked NPs can be controllers of gender agreement
in Polish. When no NP meets this requirement, the verb is in its third neuter form. In
sentences with a controller, the verb agrees with the controller.
Dyła offers a GPSG analysis of these constructions. He analyzes the NP-z-NP string as
an instance of coordination of two NPs, one of which looks like a PP. He makes this assumption on theory internal grounds. On his analysis, masculine is a default value for the
feature [GENDER] on a human or animate NP. The value for [GENDER] of the NP in the
z-NP string contributes to the value for [GENDER] on the mother node, otherwise it will
be impossible to provide a systematic account of gender resolution in quasi comitatives.
In order to make this idea work, Dyła treats z neither as a conjunction, nor as a full-fledged
preposition. The z-phrases are rather NPs with proclitic prepositions. He provides a number
2
Here and in the further discussion of McNally (1993), I am using the term ’comitative coordination’
instead of s-coordination to preserve the original terminology of the authors.
of arguments to support this claim. Polish PPs, for instance, behave very much like NPs
with respect to extraction. Both NPs and PPs allow extraction of left branches. In other
words, Polish allows extraction of prepositions and determiners out of what look like PPs
as if they were single constituents.
(18) Z
ktora
rozmawiałes dziewsczyna?
with which.Instr talked.2SG girl.Instr
lit.: With which did you talk girl?
Dyła argues for the following internal structure of the so-called PP in Polish:
NP
NP
NOM
N
P[+CL]
N[-CL]
(19)
The NP-z-NP string does not involve a PP that could be a barrier to inheritance of gender. In
order to account for the fact that the preposition z co-occurs with the NP in the instrumental
case, Dyła considers two possible assumptions 1) z subcategorizes for an instrumental NP,
or 2) that z governs the instrumental case on N. For the former, the preposition would have
to be a sister of the instrumental NP. For the latter, no such assumption is needed. Dyła’s
solution to this problem is that z governs the instrumental case on N (pp.408-410):
(20) FCR4: [P[PFORM z] [GOV N[CASE INSTR]]]
In addition, Dyła provides further constraints to account for a wider range of data in Polish.
He insures that the NP in quasi-comitative coordination must be a proper noun and never a
pronoun:
(21) FCR5:NP[CONJ NIL,FIRST P[PFORM z] [-COMMON,-PRO]
(22) CCR19: [NP[CONJ NIL,-COMMON,-PRO,FIRST P[PFORM z]] NP[CONJ
NIL,-COMMON,-PRO]]
Dyła’s multi-headed analysis perfectly accounts for the number and gender agreement facts
in Polish. However, there are several factors that seem to suggest we might want look for
another way of explaining the comitative phenomenon in Russian. First, the agreement
facts in Russian are much more simpler: they do not involve gender. In addition, there is
evidence in Russian that suggests that s-coordination and ordinary coordination are different syntactic structures, contrary to Dyła’s assumption:
S-coordinate constructions differ from ordinary coordination in the following ways:
1. Unlike ordinary coordination, comitatives are not iterative:
(23) (a) Daša i
Maša i
Saša.
Dasha and Masha and Sasha
Dasha, Masha and Sasha
(b) *Daša s
Mašei s
Sašei.
Dasha with Masha with Sasha
Dasha, Masha and Sasha
Mašei i
Petja s
Sašei.
(c) Daša s
Dasha with Masha and Petya with Sasha
Dasha and Masha, Petya and Sasha
2. S-coordination is limited to NPs, as opposed to i that has essentially the same properties as does and in English:
(24) (a) *Vanja kupil s
pročital knigu.
Vanya bought with read
book
Vanya bought and read the/a book.
(b) *Vanja čital knigu včera
s
segodnja.
Vanya read book yesterday with today
Vanya was reading a/the book yesterday and today.
(c) *Vanja krasivyi s
umnyi paren’.
Vanya handsome with
smart guy
Vanya is a handsome and smart guy.
(d) *Vanya prochital knigu s
ja napisala pis’mo.
Vanya read
book with I wrote
letter
Vanya read a/the book and I wrote a/the letter.
3. Comitatives behave differently with possessive pronouns McNally (1993). In Russian there are both reflexive and non-reflexive pronouns; only the latter may be used
when a possession relation holds between the entities denoted by the coordinate
phrase. McNally (1993) claims that both kinds of possessive pronouns can appear
in comitative constructions. However, native speakers of Russian that I consulted
accept only comitative constructions with reflexive pronouns (where the NP subject
is coindexed with the possessive that modifies the PP):
ego semja pereexali v Moskvu.
(25) (a) Pisatel’ i
Writer and his family moved.PL to Moscow
The writer and his family have moved to Moscow.
(b) *Pisatel’ i
svoja semja pereexali v Moskvu.
Writer
and self’ family moved.PL to Moscow
The writer and his family have moved to Moscow.
ego semjoi pereexali v Moskvu.
(26) (a) *Pisatel’ s
Writer
with his family moved.PL to Moscow
The writer and his family have moved to Moscow.
svoei semjoi pereexali v Moskvu.
(b) Pisatel’ so
Writer with self’ family moved.PL to Moscow
The writer and his family have moved to Moscow.
Thus, the data from Russian seem to be different and require a different account. 3
3.2 McNally 1993
While Dyła prefers the multi-headed analysis in order to account for the number and gender agreement facts in Polish and assumes a Sag et al. (1985)-style account of number
resolution, where the instrumental case-marked NP, crucially designated a head, can share
its number and gender feature values with the mother, McNally proposes another analysis
of an analogous set of data in Russian within the same framework, GPSG. As McNally
observes, the distribution of reflexive and non-reflexive pronouns suggests that comitative
constructions and ordinary coordinates have a different syntax. The first NP in a comitative
phrase can serve as an antecedent for a reflexive possessive in the prepositional phrase,
which is impossible in the ordinary coordinate phrase. These examples are repeated below
for convenience:
(27)
(a) Pisatel’ i
ego semja pereexali v Moskvu.
Writer and his family moved.PL to Moscow
The writer and his family have moved to Moscow.
svoja semja pereexali v Moskvu.
(b) *Pisatel’ i
Writer
and self’ family moved.PL to Moscow
The writer and his family have moved to Moscow.
ego semjoi pereexali v Moskvu.
(c) *Pisatel’ s
Writer
with his family moved.PL to Moscow
The writer and his family have moved to Moscow.
3
p.c.)
As for the last two sets of data, Polish does not exhibit the same pattern with reflexive pronouns (Dyła,
svoei semjoi pereexali v Moskvu.
(d) Pisatel’ so
Writer with self’ family moved.PL to Moscow
The writer and his family have moved to Moscow.
McNally proposes the following structure for the comitative construction:
NP
NP
PP[PFORM s ]
P[PFORM s ]
NP[CASE instr]
(28)
In the structure above, an NP and PP are respectively the head and non-head daughters of
a complex NP. McNally provides the following arguments in favor of such an analysis of
comitative coordinate constructions: 1) it accounts very straightforwardly for the distribution of case morphology in the construction; 2) it establishes a parallel with the related
Plural Pronoun Construction. However, such an analysis suffers from a number of problems. First, as will be shown in the subsequent sections, Plural Pronoun Constructions
exhibit a distinct syntactic behavior and cannot be analyzed as instances of comitative coordination, so that the argument provided by McNally loses its force. In addition, having
adopted such a structure, we are left with the problem of explaining the binding and number agreement facts. Moreover, this analysis fails to predict that comitative coordination is
possible only with non-pronominal noun phrases:
(29)
(a) On s
sestroi kupil
komputer.
He with sister bought.SG.Masc.3P computer
He bought a computer with his sister.
(b) *On s
sestroi kupili
komputer.
He with sister bought.PL computer
He and his sister bought a computer.
(c) My s
druzjami kupili
komputer.
We with friends bought.PL computer
My friends and I bought a computer./We and our friends bought a computer.
In the subsequent sections, it will be shown that Plural Pronoun Constructions belong to
neither of the coordination types. However, before turning to the details of the analysis, I
will summarize the main binding facts found in Russian. This digression is necessary for
establishing an alternative syntactic analysis of s-coordination.
4
Binding Facts: Ordinary and S-Coordinate Structures
have a Different Syntax
One of the arguments for differentiating comitative and coordinate constructions comes
from the behavior of possessive pronouns. It has been shown that the first conjunct in
the ordinary coordinate phrase can bind non-reflexive possessives that modify the second
conjunct, whereas the first noun phrase in the comitative coordinate phrase binds reflexive
possessive pronouns. I do not pretend to provide even a rudimentary analysis of binding in
Russian, but I will try to summarize some important generalizations that are necessary for
the current purposes.
Russian reflexive pronouns are nominative-argument oriented, i.e. they usually have subject antecedents:
(30)
(a) Tanja
rasskazala Anje
o
sebe /nei* .
Tanya.NOM told
Anya.Dat about self.Fem.Prep/*her
Tanya told Anya about herself /*her .
(b) Tanja priglasila svoju /*ejo podrugu v kino.
Tanya invited
self’s/*her friend.fem to movie
Tanya invited her friend to go see a movie.
(c) On vstretil Tanju i
svoju /ejo /*ego sestru
He met
Tanya and self’s/her/*his
sister
He met Tanya and his /her sister.
The binding theory as formulated in Pollard Sag (1994) is not universal. Problems posed by
many languages have confirmed this. It seems that binding constraints should be languagespecific. In this paper, I will adopt the Binding Theory proposed by Kupsć Marciniak
(1996) for Polish. In English (and many other languages), possessives are analyzed as
determiners, which are single obligatory specifiers of NPs (except for mass and count plural
nouns). This approach is difficult to adopt for languages such as Russian or Polish. In
Russian, as well as in Polish, there are no obligatory determiners; in fact, they occur in NPs
relatively rarely. Besides, possessives can be mixed with other determiners. This behavior
suggests possessives should be analyzed as adjuncts rather than specifiers.
Pollard Sag (1994) allow only the constituents which are subcategorized for to be bound.
This condition seems to be too weak:
prodala svoju /*eyo mašinu.
(31) Tanja
Tanya.NOM sold
self’s/*her car
Tanya sold her car.
As the example in (31) shows, the adjunct like obligatory elements is bound by the subject.
Since adjuncts are not members of the VALENCE list, they are not subject to binding
constraints. Kupsć Marciniak (1996) propose to reformulate the level at which binding
theory is defined in order to widen the scope of the theory. The idea that the binding theory
should account both for adjuncts and circumstantials, i.e. optional constituents, should be
considered as well.
VALENCE attributes are independently included in the VALENCE structure that contains
SUBJ(ECT) and COMP(LEMENT)S. Kupsć Marciniak (1996) omit the SPR attribute since
they analyze specifiers as adjuncts. The subcategorization principle is fulfilled by valence
attributes, whereas ARG–ST remains unaffected and serves mainly as the locus of the
binding theory. Following Pollard Calcagno (1997) and Kupsć Marciniak (1996), I assume
that ARG–ST should not be a list (unlike in Manning Sag (1999)) and that ARG–ST should
have a richer internal structure with attributes of its own. So, I adopt the following structure:
(32)
category
HEAD head
valence
VALENCE SUBJ synsem-list
COMPS list-of-synsems
arg-st
SUBJ synsem-list
ARG-ST COMPS list-of-synsems
ADJ list-of-synsems
CIRC list-of-synsems
The distinction between VAL SUBJ and ARG–ST SUBJ values will be crucial for the
binding theory. The ADJ attribute corresponds to ADJUNCTS that modify NPs, while
CIRC(UMSTANTIAL) represents modifiers of VPs. This distinction is useful since both
(CIRC and ADJ) modify different kinds of phrases. Following Kupsć Marciniak (1996), I
assume that possessives are elements of the ADJ list, i.e. they are treated as adjuncts. 4
The main differences between the binding theory that is adopted and Pollard Sag (1994)
are the following:
(33)
4
(1) This binding theory is based on the property of being a subject and not on the
obliqueness relation;5
Presumably, not all adjuncts will appear at the ARG–ST, but only those that are relevant for binding and
extraction. Since this issue goes beyond the scope of this paper and irrelevant for the analysis of comitatives,
I leave it open.
5
Kupsć Marciniak (1996) provide examples that show that the linear order of constituents should also
be taken into account when considering the distribution of personal and possessive pronouns- an additional
piece of evidence that the obliqueness relation is not sufficient.
(2) The theory is defined in terms of the ARG–ST structure, which contains
optional elements as well, not at the VALENCE level. 6
The relation corresponding to the o-command of Pollard Sag (1994) is called subject command (s-command):
(34) Let Y and Z be synsem objects with distinct LOCAL values. Then Y s-commands Z
just in case either:
(i) Y belongs to the ARG–ST SUBJ list, and Z belongs to one of the ARG–ST
lists; or
(ii) Y s-commands some X and Z belongs to one of X’s ARG–ST lists; or
(iii) Y s-commands some X that is a projection of Z.
The definition of local s-command is very similar to the definition of s-command. The only
difference is that in the former definition recursion is stopped when the head of a phrase
has a non-empty subject at the ARG–ST level. The definition of local s-command is given
below:
(35) Let Y and Z be synsem objects with distinct LOCAL values. Then Y locally
s-commands Z just in case either:
(i) Y belongs to the ARG–ST SUBJ list, and Z belongs to one of the ARG–ST
lists; or
(ii) Y locally s-commands some X and Z belongs to one of X’s ARG–ST lists; or
(iii) Y localy s-commands some X that is a projection of Z and Z’s value of
ARG–ST SUBJ is an empty list.
The definition of local s-command cannot be formulated (unlike in English) without recursion, which is necessary for the proper analysis of preposition phrases and sentences with
embedded noun adjuncts:
(36) Tanja prodala mašinu zheny
eyo* /
Tanya sold
car.Acc wife.Gen her
Tanya sold a car of her brother’s wife.
6
brata.
brother.Gen
Kupsć Marciniak (1996) modify the nominal object hierarchy to account for Polish facts. Since this issue
is not directly relevant for the specific data I want to account in this paper, I will not provide the full details
of their analysis.
The pronoun in (36) should be considered locally, but it is not at the ARG–ST level of the
sentence head. It appears deeper in the NP structure and for this reason the recursion (with
the stop condition added) is necessary.
The definition of local and global o-binding and o-free relations do not need reformulation;
for ease of presentation, we will rename them: (local) s-binding and (locally) s-free.
(37) Y (locally) s-binds Z just in case Y and Z are coindexed and Y (locally)
s-commands Z. If Z is not (locally) s-bound, the it is said to be (locally) s-free.
Following the proposal for Polish, I can now formulate the binding theory for Russian.
(38) BINDING THEORY FOR RUSSIAN7 :
Principle A. A reflexive anaphor must be s-bound.
Principle B. A pronoun must be locally s-free.
Principle C. A non-pronoun must be s-free.
Now consider (31) again. This is a simple sentence with a possessive expression. svoju
(’self’s’) is a possessive anaphor and should be s-bound according to principle A. Tanya is
the s-binder of svoju in case they are coindexed and Tanya s-commands the phrase svoju
mašinu because Tanya is the subject and the phrase svoju mašinu is a complement, (i.e.
belongs to the ARG–ST) of the same lexical head, prodala. According to clause (iii) of
the definition of (34), if Tanya s-commands the whole phrase, it s-commands its head, in
this case mašinu. Svoju is an adjunct of mašinu and therefore belongs to the ARG–ST of
mašinu. Applying clause (ii) of (34) I prove that Tanya s-commands svoju.
5
Formal Analysis of S-coordination of NPs
As evident from the data discussed so far, the s- and ordinary coordinate constructions
should be treated separately in Russian. In addition, as has been pointed out in the first
section, only non-pronominal NP’s can participate in s-coordination:
(39)
7
(a) On s
sestroi kupil
komputer.
He with sister bought.SG.Masc.3P computer
He bought a computer with his sister.
The principles of the binding theory are not applied if VAL—SUBJ is not a noun phrase for the head of
the phrase
(b) *On s
sestroi kupili
komputer.
He with sister bought.PL computer
He and his sister bought a computer.
I propose to treat s as a nominal head that selects two noun complements.
(40)
PHON s HEAD noun
VALENCE
SUBJ NP COMPS NP CONTENT INDEX NUM plu
The s-coordinate structure will look like the following:
NP HEAD
VALENCE
CONTENT
HEAD
VALENCE
CONTENT
IND i
RESTR
RESTR
HEAD
IND i+j
NP HEAD
npro HEAD
COMPS
VALENCE COMPS
SUBJ
NP
npro HEAD
NP
CONTENT
NP
SUBJ
VALENCE
CONTENT
npro instr
NP
CONTENT
noun
CASE instr
VALENCE HEAD
IND j
RESTR
(41)
In an s-coordinate structure, the nominal head s subcategorizes for a subject NP and a
complement NP . Moreover, the head values of the nominal head are token identical to the head specification of its subject as represented in the avm above. It is important
to notice that this head selects two NP’s whose heads are non-pronominal.
One might argue that such a treatment of s is counter-intuitive. However, unless there are
facts that argue against this analysis, I see nothing wrong with it. Moreover, such an analysis does not contradict any generalizations made about what cases Russian nouns assign
to their subjects and/or complements. Generally, Russian nouns can take bare instrumental NP complements, e.g. upravlenie predprijatiem(’administration [of] company.Instr’),
uvlečenie lingvistikoi (’passion [for] linguistics.Instr’) and so on.8 . The Russian s does not
inflect and cannot be possessed. But it exactly the properties of some Russian nouns: pal’to
or kofe, for example, do not inflect in Russian; and they cannot appear in the possessive
form either.
Such an analysis has several immediately appealing aspects. First, it insures the right number specification of the mother NP. Second, it predicts that when a possession relation
holds between the entities denoted by the comitative coordinate NP, a reflexive pronoun is
required. Third, the first NP always varies in case in accord with the grammatical relation
it bears, while the second is invariably in instrumental case. Moreover, it predicts that only
non-pronominal NP’s will participate in s-coordination and provides grounds upon which
to explain constrasts between comitative coordinates and ordinary coordinates on the one
hand, and between comitative coordinates and Plural Pronoun Constructions on the other.
Consider the following example:
svoei sestroi eli pirog.
(42) Sveta so
Sveta with self’s sister ate pie
Sveta and her sister were eating the pie.
In this example, since Sveta is the subject and svoei sestroi is a complement of the same
nominal head s, Sveta is the s-binder of svoju and Sveta s-commands the phrase svoju sestru. According to clause iii of (34), if Sveta s-commands the whole phrase, it s-commands
its head sestroi. Svoei is an adjunct of sestroi and therefore belongs to the ARG–ST of
sestroi. Applying ii of (34), one can see that Sveta s-commands svoei. Thus, keeping in
mind the binding theory proposed in the previous section and defining the structure for
comitative construction as in (41), we can account for the peculiar behavior of comitative
coordination.
8
It is interesting to notice that the cases of noun-into-preposition or preposition-into-noun conversion have
been registered in different languages. See Longobardi (2001) for the reanalysis of Latin casa ’(at) the home
(of)’ as a preposition in various Romance languages.
One additional point that we need to make here is that comitative coordinate constructions
are not iterative. In order to ensure the non-iterative behavior of these constructions, it
is possible to modify a nominal object hierarchy, presented in Pollard Sag (1994), in the
following way:
(43)
nom obj
pro
npro
non com nom
com nom
Now, any comitative coordinate NP will be of type npro and its syntactic arguments are of
type non-com-nom, thus blocking the iteration.
The question, then, is how we block the occurrence of reflexive pronouns when a possessive
relation holds between the entities denoted by the ordinary cooordinate NP. One approach
to ordinary coordination would be to accept the central claim of the coordination analysis
in Sag et al. (1985) and the analogous treatment of Pollard Sag (1994). Although such
an approach is highly problematic for coordination of unlikes and neutralization, it will be
sufficient for the current purposes. I will treat the conjunction i as a marker. Thus, ordinary
binary coordination will be a coordinate structure whose daughters are an UNMARKED–
CONJ–DTR and a MARKED–CONJ–DTR. The tree and the avm below illustrate the idea:
(44)
HEAD noun
UNMARKED-CONJ-DTR NP NUM DTRS
NP NUM MARKED-CONJ-DTR MARKING
i
(45)
NP HEAD NP HEAD NP
HEAD MARKING Maša
MRKR HEAD mark SPEC
MARKING
NP HEAD i
Sveta
i
Applying this structure together with the Binding Theory we adopted in this paper, we can
block the occurrence of the reflexive pronoun in the second conjunct:
(46) *Sveta i
svoja sestra eli pirog.
Sveta and self’s sister ate pie
Sveta and her sister were eating the pie.
In (46), the coordinate NP Sveta i svoja sestra will belong to the SUBJ list of the same head
eli, i.e. svoja is locally s-free here.9
6
Plural Pronoun Constructions (PPC’s)
This section discusses an additional kind of data, namely Plural Pronoun Constructions
(PPC’s).
Consider the following example:
(47) My s
Dašei kupili
komputer.
We with Dasha bought.PL computer
We bought a computer with Dasha/ Dasha and I bought a computer.
9
It is interesting to notice that the conjunction ili (or) is not used for nominal coordination and thus,
unlike English, the problem with having a non-homogeneous agreement with verbs (either singular or plural,
depending on the number specifications of the second conjunct) does not arise at all.
The construction in (47) is known as the Plural Pronoun Construction (PPC). 10 The PPC
has been analyzed as an instance of comitative coordination (as in Aissen (1988), Aissen
(1989)) or as adjunction Ladusaw (1989). However, as has been observed in Ladusaw
(1989) and Vassilieva (2000), PPC’s possess syntactic and semantic properties that are in
contrast with those exhibited by coordination. In this section, PPC’s are compared to other
comitative constructions and shown not to constitute a homogenous class with them: PPC’s
that have a non-inclusive interpretation are adjuncts, whereas PPC’s that have an inclusive
interpretation belong to neither adjunction nor coordination.
In Russian, the PPC is ambiguous with respect to the number of people who participate in
an event, as indicated in the translation of (47). The next two sections summarize the main
arguments(see Vassilieva (2000) among others) in favor of distinguishing the two types
of PPC’s. The two interpretations of PPCs are non-inclusive and inclusive interpretations.
The former involves the referents denoted by the plural pronoun and the comitative element
together, and the latter the intepretation when the denotation of the instrumental phrase is
included in the denotation of the plural pronoun.
6.1 Non-inclusive PPCs (NIPPC’s)
A number of tests have been discussed to determine whether a comitative phrase is a conjunct or an adjunct. One of the tests was the ability to bind reflexives. Consider the following sentence:
(48) [non-inclusive]
Dašei prodali svoi /* /* + komputer.
My s
We with Dasha sold.PL self’s /* /* + computer
We sold our /* /* + computer with Dasha .
On the non-inclusive interpretation, the antecedent of the possessive reflexive svoi is understood as my ’we’, without the inclusion of Dasha. This fact suggests that the s-phrase
in (48) must be an adjunct not a conjunct. (Note, however, (48) can have the inclusive
reading.)
When a non-inclusive PPC cooccurs with the verbs such as believe which trigger only a
distributive reading, the intended interpretation is impossible.11
10
11
Note that in (47), the use ja (I), instead of my (we) is impossible.
See Vassilieva (2000) for additional data
(49) [non-inclusive]
My s
Dašei verim
v boga.
We with Dasha believe.PL in God
We believe in God with Dasha.
(This sentence is perfectly sensible, however, on the inclusive interpretation).
Additional evidence that the non-inclusive PPC does not have a distributive reading is the
fact that it never cooccurs with recipocals:
(50) [non-inclusive]
My s
Dašei ljubim drug druzhku
We with Dasha like
each other
We like each other with Dasha
(This sentence is fully acceptable on the inclusive interpretation).
Additional evidence comes from wh-extraction. We have seen that comitative adjuncts can
undergo wh-extraction, whereas comitative conjuncts cannot. Consider the following:
(51) [non-inclusive]
S
kem
my kupili
komputer?
with whom we bought.PL computer
With whom did we buy the computer?/ NOT: Who is is such that that person and I
bought the computer?
Non-inclusive PPCs, then, cannot be analyzed as instances of coordination.
Remember that with-conjuncts cannot appear in postverbal positions. However, consider
the following:
(52) [non-inclusive]
My kupili
s
Mašei komputer.
we bought.PL with Masha computer
We bought a computer with Masha.
The examples above show that non-inclusive PPCs do not behave like comitative coordinate
constructions; their syntactic behavior resembles comitative adjuncts rather than comitative
conjuncts. So, my conclusion is that PPCs are not an instance of coordination, but instead
of comitative adjunction on the non-inclusive interpretation.
6.2 Inclusive PPCs (IPPC’s)
In this section, I consider the “inclusive” interpretation of PPCs, i.e. when the denotation
of the instrumental phrase is included in the denotation of the plural pronoun. However,
this is not the only difference.
The elements of the inclusive PPCs cannot be conjoined by i(and), unlike coordinate comitatives, where such a substitution is allowed.
(53)
(a) [ordinary coordination]
My [i
Maša] kupili
komputer
We [and Masha] bought.PL computer
i. Masha and we bought a computer.
ii. *Masha and I bought a computer.
(b) [inclusive PPC]
My [s
Mašei] kupili
komputer.
We with Masha bought.PL computer
i. *Masha and we bought a computer.
ii. Masha and I bought a computer.
Comitative coordinate constructions trigger plural agreement with the verb. The order of
conjuncts can be freely changed in both comitative and regular coordination:
(54)
(a) Maša s
Dašei(=Daša s
Mašei) pokupajut
komputer.
Masha with Dasha(=Dasha and Masha) buy.Pres.3PL computer
Masha and Dasha are buying a computer.
(b) My s
Dašei pokupaem
komputer.
We with Dasha buy.Pres.1PL computer
Dasha and I are buying a computer.
(c) *Dasha s
nami pokupaem/pokupajut komputer.
Dasha with us
buy.Pres.1PL/3PL
computer
Dasha and we are buying a computer.
As evident from (54b) and (54c), plural agreement is impossible in inclusive PPCs when
the order of the elements of the comitative phrase is altered. Moreover, pronouns cannot
participate in s-coordination in Russian:
(55)
(a) *On so
mnoi delaem/delajut
uroki.
He with me do.Pres.1PL/3PL homework
He and I are doing the homework assignment.
(b) ?On i
ja delaem
uroki.
He and I do.Pres.1PL homework
He and I are doing the homework assignment.
As has been noticed by Urtz (1994), native speakers tend to avoid and-coordination of
pronouns (as in (55b)), giving preference to PPCs (cf.(56)) as ’more natural’:
nim delaem
uroki.
(56) My s
We with him do.Pres.1PL homework
He and I are doing the homework assignment.
In addition, there is a restriction on the order of elements in IPPCs with respect to the person
hierarchy. While English requires the 1st person pronoun to come second in a coordinate
structure (He and I vs. *I and he), Russian has no preference. However, in inclusive PPCs,
the plural pronoun must be higher in the hierarchy ( ) than a (non)-plural pronoun
in the with-phrase. Therefore, the example below cannot have an inclusive reading even
when the verb agrees with the plural pronoun in person:
(57)
(a) My s
nei pojom.
we with her sing.1PL
She and I sing.
(b) *Oni so
mnoi pojom/pojut.
They with me sing.1PL/3PL
She/he and I sing.
Thus, the strict selectivity imposed on the constituents and their order is unique to IPPCs
and does not apply to other comitative constructions. This supports the hypothesis that the
IPPC is not an instance of coordination.
Additional evidence comes from the tests that we used to distinguish comitative conjuncts
from adjuncts. With respect to these tests, IPPCs pattern with neither of these constructions. In some aspects, IPPCs resemble adjuncts: they allow the verb to separate its complements.12
(58) My kupili
s
Anej komputer.
We bought.PL with Anya computer
Anja and I bought a computer.
12
Note that (57b) and (59b) are grammatical under the adjunct interpretation.
In other aspects, inclusive PPCs resemble comitative coordinate constructions. They behave similarly with respect to wh-extraction (59) and the ability to antecede reflexives (60),
and can appear with reciprocals and verbs that require a distributive reading ((62),(61)).
(59)
Anej učilis.
(a) My s
we with Anya study.PL
Anya and I were studying.
(b) *S
kem
my učilis’?
With whom we study.PL
With whom were we studying?
Anei čitaem
knigu.
svoju* ,* , +
(60) My s
We with Anya read.1PL self’
book.
Anya and I are reading our book.
(61) My s
Anej ljubim drug druga
We with Anya like.PL each other
Anya and I like each other
(62) My s
Anei znaem
kitaiskii.
We with Anya know.1PL Chinese
Anya and I know Chinese.
It has been shown that IPPCs share some properties with comitative adjuncts and comitative conjuncts. In addition, they exhibit a number of characteristics not found in other
with-constructions. We assume that they are syntactically neither instances of adjunction
nor of coordination. The next section provides a formal analysis of IPPCs as comitative
complementation.
7
Formal Analysis of PPC’s
As has been argued above, the with-phrase in non-inclusive PPC’s is an adjunct. Thus, no
special attention is required to these constructions.
However, we have shown in the previous sections that IPPCs share some properties with
comitative adjuncts and comitative conjuncts. Like coordinate comitative constructions,
IPPCs trigger plural agreement with the verb, jointly antecede reflexives, prohibit ’whextraction’, can appear with reciprocals and distributive verbs; however, unlike coordination and like adjunction, the prepositional phrase can occur in postverbal position. The
most intriguing fact is that they exhibit a number of features not found in other withconstructions: the order of the constituents with respect to their person specification is
unique; an IPPC cannot be substituted with an equivalent coordinate phrase. Since IPPCs
pattern neither with coordinate comitatives nor with comitative adjunction, we assume that
the plural pronouns that occur in such structures are heads that select with-phrases with
particular specifications. Such an assumption requires some additional modification of the
hierarchy of nominal objects. According to this hierarchy, pronouns can be transitive and
intransitive. In Russian, transitive prounouns are plural and select a PP complement, and
therefore, will be treated as transitive13 :
(63)
nom obj
npro
non com nom
pron
com nom ana
ppro
pro i
pro tr
(64)
PHON phon HEAD noun
CAT
VAL COMPS PP
CONTENT 13
INDEX Notice that transitive pronouns exist in some languages, e.g. Cayuga, an Iroquoian language.
(65)
NP HEAD H
C
HEAD
NP PP HEAD VAL COMPS PP H
C
my
HEAD
P
prep PFORM s
VAL COMPS
NP
nei
s
(64) and (65) above illustrate the idea. The plural pronoun selects a PP . However, such
an account is not complete. As has been discussed in the previous sections, the denotation
of the complement should be included in the denotation of the head (in order to get the
inclusive interpretation) and also, we must impose restrictions on the order of elements in
IPPCs with respect to the person hierarchy. We can then define the following relation:
(66) include–in(SUPERSET, SUBSET) iff SUBSET is a subset of SUPERSET and the
cardinality of the SUPERSET denoted by the plural pronoun is one greater than the
cardinality of the SUBSET denoted by the object of s and the denotation of the
SUBSET is included in the denotation of the SUPERSET.
Perhaps the best approach to this phenomenon is to assume that there is a pragmatic constraint requiring that the relation holds between the elements of the IPPC and state another
constraint on the person hierarchy. We assume that the person hierarchy is a linear order on
person values and we state a disjunction as exemplified in (67): the possible person values
are expressed as an object language relation (and not as a pragmatic constraint).
(67)
PHON phon CAT
VAL COMPS PP
CONX BG
( PERS 1rst
PERS 3rd)
include–in
SUPERSET
SUBSET PERS(2nd
CONTENT pro tr INDEX HEAD noun
3rd))
( PERS 2nd
PERS 3rd) ( PERS 3rd
Having defined such a lexical entry for the transitive plural pronoun, we can easily account
for the inclusive plural pronoun phenomenon. Using the include-in relation makes the
correct prediction that the first conjunct in the inclusive PPC has to be plural.
8
Problems and Issues for Further Research
This paper began by considering two types of comitatives: adjuncts and coordinate constructions. The evidence to distinguish the two types of structure comes from agreement
patterns, reflexivization, discontinuity effects and semantic interpretation. The next concern of the paper was to differentiate between the two types of nominal coordination in
Russian: s- and ordinary coordination. The argument in favor of this distinction comes from
the difference in the behavior of these constructions with respect to possessive pronouns
when the possession relation holds between the entities denoted by the ordinary/comitative
coordinate phrase. Two distinct structures were proposed: 1) the nominal head s selects
two NP complements in the case of comitative coordination; and 2) i(and) is a marker
in ordinary coordinates. We followed the standard HPSG analysis and used a coordinate
structure whose daughters are UNMARKED–CONJ–DTR and MARKED–CONJ–DTR.
The paper discussed the Plural Pronoun Constructions as well. It has been shown that
PPCs cannot be accounted for by a single syntactic structure. Some of them are regular
comitative adjuncts (NIPPCs), whereas the others (IPPCs) can be ascribed to neither of
the two syntactic types: in some cases their syntactic behavior resembles adjunction (they
allow verbs to separate their complements); in some other cases they resemble comitative
coordination (wh-extraction, ability to antecede reflexives, plural agreement and coocurrence with reciprocals). However, unlike comitative coordination, the elements of IPPCs
cannot be conjoined by i; plural agreement is impossible when the order of the elements
of IPPC’s is altered. In addition, there are restrictions on the order of elements in IPPCs
with respect to the person hierarchy; the denotation of the head pronoun should include the
denotation of the complement it selects.
However, there are several issues with the analysis proposed in this paper. Althought the
majority of native speakers of Russian do not accept the construction in (68b), some speakers do (even though they usually prefer (a) over (b)):
(68)
(a) Tanja s
Vanej
i
Petej
Tanya with Vanya.Instr and Petya.Instr
Tanya with Vanya and Petya
(b) Tanja s
Vanej
i
s
Petej
Tanya with Vanya.Instr and with Petya.Instr
Tanya with Vanya and with Petya
The sentence in(68b) is not a counterexample to the account proposed in this paper.The
example in (68b) can be analyzed as an instance of left node raising, comparable to English right node raising, where both [s Vanej] and [s Petej] share their respective remaining
valent, analogously to the English Mary’s photograph of John and picture of Susan 14 .
In addition, we provided only a tentative binding theory for Russian. The further investigation will have to account for a wider range of data, such as reciprocal pronouns, emphatic
reflexives (e.g. sam sebja ’self own’) etc.
A linearization account in a relatively free word order language, such as Russian of the
fact that s- (as well as ordinary) coordinate NP’s do not allow any discontinuities and that
IPPC’s do not extract but allow the verb to separate its complements should be provided.
Acknowledgments
I would like to thank Nick Cipollone, David Dowty, Richard Janda, Brian Joseph, SunHee Lee and Detmar Meurers. Special thanks to Bob Levine, Rick Nouwen, Carl Pollard,
Nathan Vaillette and Shravan Vasishth for reading earlier versions of this paper and providing valuable comments. All mistakes are mine.
References
A ISSEN , J UDITH L. 1988. Agreement Controllers and Tzotzil Comitatives. Language
510–536.
14
Thanks to Bob Levine for pointing this to me
A ISSEN , J UDITH L. 1989. Agreement Controllers and Tzotzil Comitatives. Paper presented at the LSA Annual Meeting in New Orleans.
C AMACHO , J OSE. 1996. Comitative Coordination in Spanish. In Aspects of Romance
Linguistics, ed. by Claudia Parodi et al., 107–122. Washington D.C.: Georgetown University Press.
DYŁA , S TEFAN. 1988. Quasi-comitative Coordination in Polish. Linguistics 26.383–414.
K ATHOL , A NDREAS. 2000. Linear Syntax. Oxford University Press.
K UPS Ć , A NNA, M AŁGORZATA M ARCINIAK. 1996. Some Notes on HPSG Binding Theory for Polish. In European Conference on Formal Description of Slavic Languages,
pp.193–203.
L ADUSAW, W ILLIAM. 1989. Group Reference and the Plural Pronoun Construction. Paper
presented at the LSA Annual Meeting in New Orleans.
L ONGOBARDI , G IUSEPPE. 2001. Formal Syntax,Diachronic Minimalism, and Etymology:
The History of French Chez. Linguistic Inquiry 32.275–302.
M ANNING , C HRISTOPHER D., I VAN A. S AG. 1999. Dissociations Between Argument
Structure and Grammatical Relations. In Lexical and Constructional Aspects of Linguistic Explanation, ed. by Andreas Kathol Gert Weberlhuth, Jean-Pierre Koenig, 63–78.
CSLI.
M ARY DALRYMPLE , I RENE H AYRAPETIAN, T RACY H OLLOWAY K ING. 1998. The
Semantics of the Russian Comitative Construcions. Natural language and linguistic
theory 597–631.
M C N ALLY, L OUISE. 1993. Comitative Coordination: A Case Study in Group Formation.
Natural Language and Linguistics Theory 11.347–379.
P OLLARD , C ARL, M IKE C ALCAGNO. 1997. Argument Structure, Structural Case, and
French Causatives. HPSG97, Cornell University.
P OLLARD , C ARL, I VAN A. S AG. 1994. Head-Driven Phrase Structure Grammar. The
University of Chicago Press, Chicago and London.
S AG , I VAN, G ERALD G AZDAR, T HOMAS WASOW, S TEVEN W EISLER. 1985. Coordination and How to Distinguish Categories. Natural language and linguistic theory
3.117–171.
S CHWARTZ , L INDA. 1988. Asymmetric Feature Distribution in Pronominal ”Coordinations”. In Agreement in Natural Language, ed. by Michael Barlow Charles A. Ferguson,
237–249. CSLI.
S CHWARTZ , L INDA. 1988. Conditions for Verb-coded Coordinations. In Studies in Syntactic Typology, ed. by Edith Moravcsik Michael Hammond Jessika Wirth,
Typological Studues in Language (TSL), 53–73. John Benjamins Publishing Company.Amsterdam/Philadelphia.
S TASSEN , L EON. 2000. AND-languages and WITH-languages. Linguistic Typology 4.1–
54.
U RTZ , B ERNADETTE J EAN. 1994. The Syntax, Semantics, and Pragmatics of a Nominal
Conjunction: The Case of Russian i. Harvard University dissertation.
VASSILIEVA , M ASHA. 2000. On the Typology of Russian Comitatives. In Proceedings of
FASL:Bloomington Meeting.Slavica Publishers.
V ASSILIEVA , M ASHA, R ICHARD K. L ARSON. 2001. The Semantics of the Plural Pronoun
Construction. Paper presented at the Semantics and Linguistics Theory XI (SALT 11),
NYU, New York, NY.